Activity-Evoked Increases in Extracellular Potassium Modulate ...
JOURNALOFNEUROPHYSIOLOGY Vol. 58, No. 2, August 1987. Printed
in U.S.A.
Activity-Evoked Increases in Extracellular Potassium Modulate Presynaptic Excitability in the CA1 Region of the Hippocampus N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
DepartmentofNeuroZogy,Stanford University Schoolof Medicine, Stanford, California 94305; Yale Medical School,New Haven, Connecticut06510;and Palo Alto VeteransAdministration Medical Center,Palo AZto,Calzfornia 94304
to 7 mM, asrecorded in the stratum radiatum with potassium ion-sensitive microelectrodes. 1. The effects of stimulus-evoked potassium When postsynaptic activity was blocked, acreleaseon the excitability of presynaptic axons tivity-evoked rises in [K+], were reduced to were studied in the rat hippocampal slice t25% of their former value. This suggeststhat preparation. Extracellular stimulation and re- activity-evoked increasesin [K’j, derive precording in the stratum radiatum of CA1 dominantly from postsynaptic elements. yielded a characteristic field potential corre5. Superfusion of solutions containing elsponding to the compound action potential of evated [K+] produced biphasic changesin the nonmyelinated afferents and subsequent excitability of CA1 afferents that were qualipostsynaptic activation of pyramidal cells. tatively similar to those produced by repetitive 2. Repetitive stimulation ( 1 s; 2- 100 Hz) stimulation. Elevated [K’10 below 6 mM proproduced biphasic changes in the excitability duced increased excitability, whereas [K’10 of the afferents. Initial responsesshowed in- above 6 mM yielded decreasedexcitability. creased conduction velocity and variably in6. These results demonstrate that in the creased amplitude; subsequent responses CA1 region of the hippocampus, significant showed progressively decreasing conduction rises in [K’10 occur with activity and derive velocity and amplitude tending toward con- predominantly from postsynaptic elements. duction block. Decreasesin excitability were The conduction properties of CA1 afferents maximal at the end of stimulation and were are sensitive to the level of [K’lO, whether almore pronounced with higher stimulation tered artificially or by activity. These effects frequencies. may constitute a mechanism of postsynaptic 3. When synaptic transmission was abol- modulation of presynaptic conduction operished with superfusate containing elevated ating within a broad range of afferent firing [Mg2’] (6 mM) and decreased [Ca2’] (0.25 frequencies in the hippocampus. mM), kynurenic acid (1 mM), or adenosine ( 100 PM), the ability of the fibers to follow INTRODUCTION repetitive stimulation was enhanced, as indicated by a reduction in amplitude decrement The transmembrane potassium gradient is of the presynaptic volley. The decreasein con- the principal determinant of resting potential duction velocity at the end of stimulation was and therefore can profoundly influence neulessthan half that obtained with intact post- ronal excitability ( 1, 11). During neuronal activity, relatively large increasesin extracellular synaptic activity. 4. Concomitant with changes in the excit- potassium concentration ([K+],) can occur, ability of CA1 afferents, the concentration of yielding depolarization of surrounding neuextracellular potassium ([K+],) increased up rons. This results both from an abundance of SUMMARY
404
AND
CONCLUSIONS
0022-3077/87
$1 SO Copyright
0 1987 The American
Physiological
Society
EXTRACELLULAR
POTASSIUM
potassium conductances that are activated during neuronal activity and from a restricted extracellular space in which small activityevoked effluxes of K+ produce significant changes in [K’10 (25). Because of the close packing of neuronal elements in the CNS, the possibility exists that activity of one set of neurons could affect excitability within the same or separate sets of neurons via depolarization resulting from changes in [K+],. Elevations in [K’10 have been observed in pathological states in the central nervous system (for a review, see 3 1) such as seizure discharge (30) and spreading depression (18,26). Significant rises in [K’10 have also been observed following physiological levels of activity in spinal dorsal horn (17) cerebellar cortex (26) retina (8) and hippocampus (3,6). However, it has been difficult to establish a functional role for the elevated [K’10 associated with either physiological or pathological states. It has been hypothesized that activity-evoked increases in [K+], underlie the primary afferent depolarization and serve as a mechanism of presynaptic inhibition ( 17). The excitability of the presynaptic parallel fibers of the cerebellar cortex has been found to be reduced during stimulus-evoked elevation of [K+], , and since much of the K+ release derived from postsynaptic elements, it was suggested that the increase in [K+], might act as a feedback mechanism to regulate parallel fiber excitability and therefore synaptic transmission ( 16,2 1). These observations give support to the notion that the extracellular ionic environment can mediate changes in neuronal excitability and thus may be regarded as a separate neuronal “communication channel” (24). In the present study, the effects of activitydependent rises in [K+], upon presynaptic axonal excitability were investigated in the in vitro hippocampal slice preparation. The results indicate that the majority of K+ efIlux during repetitive stimulation in the stratum radiatum derives from postsynaptic elements and that the resulting rise in [K’10 significantly affects the excitability of the CA1 afferents. It is suggested that modulation of presynaptic fiber excitability by accumulation of extracellular potassium from postsynaptic activity may constitute a feedback mechanism that limits afferent firing frequency and affects subsequent postsynaptic activation of hippocampal nvramidal cells.
AND
EXCITABILITY
405
METHODS
Female Wistar rats (Simonsen) were deeply anesthetized with pentobarbital sodium (60 mg/kg). After rapid exsanguination and craniotomy, a block of tissue was excised from one cerebral hemisphere using transverse cuts angled rostrally - 30° from the coronal plane. The tissue was secured to a small aluminum base using cyanoacrylate glue and buttressed against an agar block. Tissue slices 400 ,urn thick were cut using a Vibratome (Oxford Instruments), and the hippocampus was dissected free. The slices were then incubated in 34OC Krebs’ solution containing (in mM): (NaCl, 124; KC1 3.0; MgC12, 2.0; CaC12, 2.0; NaH,P04, 1.3; NaHC03, 26; dextrose, 10; pH 7.4; saturated with 95% 02 and 5% COJ for at least 1 h. Solutions with elevated [K] were osmotically balanced by subtraction of equimolar amounts of Na? During recording, slices were maintained in a submersion-type chamber with a volume of 1 ml through which Krebs’ flowed at -5 ml/min. A tungsten microelectrode (tip exposure 300 pm) was used for stimulation. Constant-current pulses controlled by a digital timing device were delivered by stimulus isolation units grounded through the bath. K+ ion-sensitive microelectrodes (IS+-ISMs), which allowed the recording of [K’lo and extracellular field potentials, were constructed from theta capillary glass (R & D Scientific Glass) using the methods described by Lux and Neher (20). Tip diameters were typically l-2 pm. The electrode barrel to serve as the K+-sensing electrode was filled at the tip with a 500~pm column of K+ ion exchange resin (Dow-Corning 4773 17) and backfilled with 2 M KCl; the reference barrel for recording field potentials contained 150 mM NaCl. The reference and K+ signals were led by Ag-AgCl wires to a highimpedance differential amplifier (Axon Instruments Axoprobe- 1). Field potentials from the reference barrel were independently amplified, low-pass filtered at 10 kHz, digitized (Nicolet 1170 and Neurodata DR-484) and stored on magnetic tape. The K+ potential was obtained by differentially amplifying the reference and K+ signals to eliminate common-mode potentials unrelated to K+ activity, low-pass filtering at 1 kHz, and monitoring on a chart recorder (Gould 220). A calibration curve for each K+-ISM was derived as a single exponential using points taken as a mean of measurements before and after experimentation from the following two test solutions (3 mM K+; 150 mM Na+) and 30 mM K+; 150 mM Na+). Calibration with a greater number of test solutions showed that the less time-consuming method allowed the estimation of [K’10 with an approximate error of t5%. Stimulating and recording electrodes were positioned in the stratum radiatum as shown in Fig. IA. Afferents in the stratum radiatum consist of the
406
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
Pl
11N2
v Nl
C
LO mV I
gc
FIG. 1. Characteristic field potentials evoked by stimulation in the stratum radiatum in the presence and absence of synaptic transmission. A: potassium ion-sensitive microelectrodes positioned in the stratum radiatum of CA1 allowed simultaneous recording of field potentials and activity-evoked extracellular K+ concentration rises following repetitive stimulation in the stratum radiatum. B: field potential recorded in the stratum radiatum resulting from a single stimulus. Nl corresponds to the presynaptic volley, whereas N2 and N3 reflect postsynaptic activity. C: field potential recorded in the stratum pyramidale resulting from a single stimulus in the stratum radiatum. The large negativity represents the pyramidal cell population spike. D-F: field potentials after blockade of synaptic transmission (arrow) by reduced [Ca2’] (0.25 mM) and elevated [Mg2+] (6 mM), kynurenic acid ( 1 mM), and adenosine ( 100 pm), respectively. Field potentials with intact postsynaptic activity are superimposed for comparison. Calibration applies to B-F.
Schaffer collaterals and commisural fibers, both of which are fine caliber, nonmyelinated axons with an average diameter of ~0.1 pm (19, 34). Activation of these axons leads to monosynaptic excitation of pyramidal cell apical dendrites. A single stimulus elicits a characteristic field potential (Fig. 1B) corresponding to the population presynaptic fiber volley and subsequent excitatory postsynaptic potential (EPSP). The initial positive-negative-positive (P lNl-P2) components of the field potential correspond to the source-sink-source currents of the propagating presynaptic volley, with N 1 representing the inward current of the action potential. Following Nl, a second negativity (N2) represents in-
ward currents associated with the population EPSP occurring within the pyramidal cell dendritic arbor (5). With increasing stimulation current, a positive component P3-N3 becomes superimposed upon N2; this component represents the source current deriving from action potentials in the pyramidal cell body, or the “population spike,” as shown by a simultaneous recording made in the stratum pyramidale (Fig. 1C). In the hippocampal slice, CA1 afferents course orthogonally to the pyramidal cell body apical dendrites (5). The stimulating and recording electrodes were positioned along the path of the afferents with a longitudinal separation of 0.1-0.5 mm to maxi-
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
2.0 msec
NS
low Ca
aden
.m v 101
kYn
FIG. 2. Effects of various synaptic blockers upon activity-dependent changes in the presynaptic volley. A: raster display of successive responses evoked by stimulation at 20 Hz for 1 s. Arrow marks first response of train. From Ze@ to right, responses were obtained in normal solution, reduced [Ca2’] (0.25 mM) with elevated [Mg2’] (6 mM), adenosine ( 100 pm), and kynurenic acid ( 1 mM). The presynaptic volley shows decreased excitability at the end of stimulation as evidenced by a decreased amplitude and increased latency. These decreases are minimal when postsynaptic activity is blocked. B: superimposed first (arrow) and last responses, which were hand traced from records of the stimulation trains in A.
mize the amplitude of the presynaptic volley (N 1) while preserving a sufficiently long conduction distance. The stimulus was adjusted to obtain a justmaximal postsynaptic (N2) response; this stimulus intensity was usually substantially submaximal with regard to the Nl response. Typically this resulted in a field potential with Nl and N2 amplitudes of 6 and 2 mV, respectively. Slices with a maximal synaptic potential of < 1 mV were discarded. Several agents were used to block synaptic activation of postsynaptic elements (Fig. 1, D-F). Increased [Mg2+] (6 mM) with reduced [Ca2’] (0.25 mM) blocks Ca2+- mediated transmitter release while minimally affecting presynaptic axon excitability. Kynurenic acid (1 mM) has been shown to antagonize the response of postsynaptic glutamate receptors at several sites in the hippocampus (12). Adenosine ( 100 PM) inhibits synaptic transmission through a putative presynaptic site of action (15, 27). Note that after disruption of synaptic transmission, only the presynaptic component of the response (P 1-N 1-P2) remained. Using synaptic blockers with diverse mechanisms of action enabled the identification of effects due to the absence of postsynaptic activity rather than to alterations in membrane properties or conductances specific to a given blocker. Excitability of an axon by definition varies inversely with its stimulation threshold (9). However, due to the impracticality of obtaining single axon recordings of fine caliber CA 1 afferents, conduction
velocity and amplitude of the population response were used as indirect measures of mean excitability. The correlation of conduction velocity and excitability has been established by single axon studies in which conduction velocity has been found to covary with threshold (29). When change in the latency of a response varies linearly with the conduction distance, latency variation can be used as an accurate measure of change in conduction velocity. Latency variation and change in conduction velocity will be used interchangeably here when that condition has been empirically met. Amplitude of a population response has also been used as a measure of excitability ( 13), and as it represents the sum of currents from synchronously activated axons, it correlates well with the number of active units (4). However, variations in resting potential will result in varying amplitudes of the constituent action currents, making the amplitude of the population response only an approximation of the number of active axons. Substantial reductions in the amplitude of the fiber volley reflect the occurrence of conduction block in a large fraction of the axon population. Evaluation of both latency variation and amplitude yields more information about excitability changes than either one alone, with latency variation giving a measure of the average excitability of a population of stimulated afferents, and amplitude approximating the size of the population.
408
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
A
A
A
050
A cl
A
A
A
0
I
0
8
"60 n 20
0
I 0
0
i A
0
0
Ofi 8A
l 1
I
I
0.4 TIME AFTER STIMULATION
0 A
A
0
1
1
0.6
I
I
0.8
I
1.0
ONSET (set)
0
8
.lO
A Cl
8
l
A
0
8
.20
0 0
A
A
A
0 0
A
20
A
050 ugj
A A
A
A
A
A80
0
1
0.0
I
I
I
0.2
I
I
0.4
I
I
0.6
TIME AFTER STIMULATION
I
OS
I
J
1.0
ONSET (set)
FIG. 3. Changes in excitability of the CA 1 afferents during repetitive stimulation with intact postsynaptic activity. Data shown in Figs. 3 and 4 are from a single representative experiment. A: change in latency of the presynaptic volley as a percentage of base-line value (first response in train) for responses recorded at various intervals during 1 s of stimulation at frequencies ranging from 10 to 80 Hz. B: amplitude of the presynaptic volley as a percentage of base-line value during 1 s of stimulation at same frequencies as in A.
RESULTS
Changes in presynapticjiber excitability produced by repetitive stimulation Activity-evoked changes in the excitability of CA1 afferents were studied by delivering trains of stimuli to the stratum radiatum. Records of successivefield potentials evoked by l-s stimulation trains at 20 Hz in normal Krebs’ solution and under conditions of synaptic blockade are shown in Fig. 24 superimpositions of the first (arrow) and last responsesof these trains are shown in Fig. 2B. In normal solution (NS), with intact postsyn-
aptic activity, significant changes in the amplitude and latency of the presynaptic volley occurred during repetitive stimulation. Two changeswere particularly evident when comparing the first and last responsesof a train: the amplitude of the last presynaptic volley had decreasedvirtually to the point of extinction, and its latency, as measured from the point of peak amplitude, had substantially increased.The large change in both of these parameters over the course of stimulation indicatesdecreasingexcitability of the presynaptic axons during repetitive stimulation. The excitability of the afferents rapidly approached
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
409
A
~80 060
I
1
1
I
I
0.6
0.4 TIME AFTER STIMULATION
ONSET (set)
A
A8O
20
0
1
0.0
I
I
0.2
I
I
I
0.4
I
0.6
TIME AFTER STIMULATION
I
I
0.8
I
1
1.0
ONSET (set)
FIG. 4. Changes in excitability of the CA 1 afferents during repetitive stimulation after abolition of postsynaptic activity with reduced [Ca”] (0.25 mM) and elevated [Mg2’] (6 mM). Data shown in Figs. 3 and 4 are from a single representative experiment. A: change in latency of the presynaptic volley as a percentage of base-line value (first response in train) for responses recorded at various intervals during 1 s of stimulation at frequencies ranging from 10 to 80 Hz. B: amplitude of the presynaptic volley as a percentage of base-line value during 1 s of stimulation at same frequencies as in A.
conduction block-that is, the population of axons responsive to a constant stimulus intensity progressively declined, and those still capable of conduction did so with decreased conduction velocity. Blockade of synaptic activation of pyramidal cell dendrites by CA1 afferents dramatically reduced the decline in excitability of the afferents during repetitive stimulation. As shown in Fig. 2, postsynaptic activity was blocked by several agents with differing mechanisms and sites of action. Regardless of the agent used, reduction of postsynaptic activity led to smaller decreases in amplitude and smaller increases in latency during repetitive
stimulation (as is evident by comparison of the first and last responses in a train) and thus greatly lessened the extent of conduction block during repetitive stimulation. Although the degree of postsynaptic blockade varied with the agent used, all were effective in restricting excitability changes of the presynaptic fibers during repetitive stimulation. When measured with single stimuli, base-line amplitude and conduction velocity of the presynaptic volley were not significantly affected by bath application of blocking agents. Quantitative comparisons of presynaptic fiber excitability during repetitive stimulation through a range of frequencies (Fig. 3) dem-
410
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
Bi 1 : :
C
k
LX
NS
2.0
40 mV msec
low ca
7
D 5
[K+1 0
(mM) 3 FIG. 5. Rises in extracellular K+ concentration ([K+],) following repetitive stimulation with intact and abolished postsynaptic activity. A: raster display of successive responses evoked by stimulation for 1 s at 20 Hz with intact postsynaptic activity. Arrow marks first response of train. B: repetitive stimulation as in A following abolition of postsynaptic activity with reduced [Ca2’] (0.25) and elevated [Mg2’] (6 mM). C: rise in [K’10 recorded during responses to stimulation shown in A. D: rise in [K’10 recorded during responses to stimulation shown in B.
onstrated the variation of excitability changes with stimulation frequency and revealed a transitory phaseof increased fiber excitability. Repetitive stimulation for 1 s at various frequencies throughout the range of lo-80 Hz produced biphasic changes in the excitability of the presynaptic fibers: several responsesfollowing the first showed decreasedlatency and constant or slightly increased amplitude of the fiber volley, whereas subsequent responses showed progressively increasing latency and decreasing amplitude. The latency variations increasedlinearly with longitudinal separation of the stimulating and recording electrodes (data not shown), thus represented changesin the conduction velocity of the fiber volley. The duration of the period of increasedconduction velocity varied approximately inversely with the stimulation frequency, but the peak magnitude of the conduction velocity increase was constant for all frequencies, - 8%. The subsequent period of decreasing conduction velocity and amplitude yielded, at the end of stimulation, a net decrease in both of these parameters from base-line value. This net decreasein excitability was observed for ah frequencies from 10 to 80 Hz, with its magnitude
varying in a complex way with frequency. The magnitudes of both increasesand decreasesin excitability showed small variability among four experiments: the mean standard deviations of measurements of latency and amplitude for all frequencies in normal solution were 5.7 and 10.9%, respectively. Both latency and amplitude variations during repetitive stimulation were greatly reduced when postsynaptic activity was blocked. Figure 4 shows the effects on excitability following blockade of postsynaptic activity by reduced [Ca2’] (0.25 mM) and elevated [Mg2’] (6 mM). Biphasic latency shifts were present, but the magnitudes of both increasesand decreases in latency were attenuated; similarly, amplitude declined over the course of stimulation, but the net change at the end of stimulation was a fraction of the value seen with intact postsynaptic activity.The differences between excitability changesobserved with and without postsynaptic activity were most marked at stimulation frequencies of 40 Hz and higher. At 40 Hz, the mean decrease in conduction velocity at the end of stimulation with blocked postsynaptic activity was 6.7 t 1.7% (SD), y2= 4, compared with a 22.9 t 4.5% decrease
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
411
.
A7
.
5
5
10
60
I
[K+1 0
(mM)
3
:[email protected]
1K+1 0
0 0
(mM)3 t
80 ,
O ,
2
10
,
0
*
20
0
,
40 STIMULATION
0
,
1
60
0
0
0
,
,
80
,
,
loo
FREQUENCY (Hz)
FIG. 6. Rises in extracellular K+ concentration ([K],) during repetitive stimulation with intact and abolished postsynaptic activity. A: rises in [K+], recorded with intact postsynaptic activity during 1 s of repetitive stimulation at frequencies ranging from 5 to 60 Hz. B: rises in [K+], recorded after abolition of postsynaptic activity with reduced [Ca”] (0.25) and elevated [Mg2’] (6 mM) during 1 s of repetitive stimulation at same frequencies as in A. Calibration of [K+], in A also applies to B. C: rises in [K+], during 1 s of repetitive stimulation at frequencies ranging from 2 to 100 Hz with intact (JiZZedcircles) and abolished (open circles) postsynaptic activity. Data shown are from a single representative experiment.
with intact postsynaptic activity; similarly, the decline in amplitude in the absence of postsynaptic activity was 3.0 t 7.2% versus 55.8 t 16.7% in the presenceof postsynaptic activity. Evaluation of paired t test statistics found significant differences between excitability changeswith intact and abolished postsynaptic activity at 20 Hz (P < 0.05) and 40-80 Hz (P c 0.001). These results were similar with blockade of synaptic transmission by either reduced [Ca”‘] (0.25 mM) and elevated [Mg2’] (6 mM) or by application of kynurenic acid (1 mM) or adenosine (100 PM) and were completely reversible in normal solution.
Change.s in [K’jo from presynaptic and postsynaptic sources Potassium ion-sensitive microelectrodes were used to correlate excitability changes of the presynaptic fibers with simultaneous changes in [K’10 in the stratum radiatum. A
single stimulus with intact postsynaptic activity evoked a rise in [K’10 of -0.1 mM, but, as will be discussedbelow, considerations of microelectrode dead spacemake this value and those obtained during repetitive stimulation likely significant underestimates of actual [K+],. Repetitive stimulation for 1 s through a range of frequencies up to 100 Hz increased [K+], up to 7 mM, a rise of 4 mM from the baseline of 3 mM. Peak stimulus-evoked [K’10 tended to occur with stimulation at 40 Hz. Decay of the rise in [K’10 occurred exponentially with a time constant of 1.1 s with intact postsynaptic activity and 1.5 s with blocked postsynaptic activity. Increasesin [K’10 above 6 mM usually decayed with an undershoot. Repetitive stimulation for 1 s at 20 Hz produced an increase in [K’10 to over 6 mM (Fig. 5, A and C). Blockade of postsynaptic activity led to a profound reduction of activity-evoked [K’lO, yielding an increase of only 0.2 mM at
412
N. P. POOLOS,
4
t
56
7
100
M. D. MAUK,
AND
J. D. KOCSIS
’ E’l, (mW 200
12 mM K+
TIME
(set)
FIG. 7. Changes in excitability of CA 1 afferents during superfusion of Krebs’ solution containing elevated [K+]. Data shown are from a single representative experiment. Osmotically balanced Krebs’ with 12 mM K+ was superfused into the bath, and single stimuli applied every 6 s. Changes in latency (Jilled circles) and amplitude (open squares) of the presynaptic volley are shown as percentages of the base-line value. Changes in extracellular K+ concentration ([K+],) over time were recorded with a K+-ISM. The initially unchanging value of [K+], reflects the transit time of the superfusate into the bath. At a time beyond that shown here, [K’10 approached 12 mM and the amplitude of N 1 approached zero; due to the decreasing magnitude of N 1 amplitude measurements above noise level, measurements when [K’10 exceeded 9 mM are not shown.
20 Hz (Fig. 5, B and D). As shown in Fig. 6A, when a range of stimulation frequencies were used in normal solution, activity-evoked K+ release increased rapidly with increasing frequency up to 40 Hz; above 40 Hz, peak [K’10 levels gradually declined with increasing stimulation frequency. Blockade of postsynaptic activity reduced activity-evoked [K+], risesto a small fraction of their former magnitude as shown in Fig. 6, B and Cusing reduced [Ca2’] and elevated [Mg2’] as a synaptic blocker. The ratio of peak [K+], with postsynaptic activity intact and abolishedwas pooled over the range of frequencies 5-80 Hz; the mean of this value over three experiments was 8.13 t 4.92 (SD). In all experiments, with stimulation of 10 Hz or above, the ratio of peak stimulus-evoked [K’10 arising from intact and abolished postsynaptic activity was at least 3: 1. This result was independent of the agent used to block postsynaptic activity and thus suggeststhat activity-evoked increasesin [K’10 in the stratum radiatum can be principally attributed to
activity of the postsynaptic elements. Since during postsynaptic blockade the observed reduction in activity-evoked [K’10 coincided with the restriction of presynaptic fiber excitability changes, the correlation between elevated [K’10 and presynaptic excitability was investigated through artificial elevation of [K+], .
Changes in CA1 aferent excitability induced by superfusion of Krebs’ containing elevated [K’jO To assess the effects of elevated [K’10 upon the excitability of the presynaptic afferents, osmotically balanced Krebs’ solution containing 12 mM K+ was superfused into the bath, causing [K’10 in the region of the recording K+-ISM to rise gradually. During the rise in [K’lO, single stimuli were delivered every 6 s and changesin latency and amplitude of the fiber volley were measured. The results of this experiment are depicted in Fig. 7. Elevating [K+], by bath superfusion induced
EXTRACELLULAR
POTASSIUM
biphasic changes in the conduction properties of CA 1 afferents which were qualitatively similar to those induced by repetitive stimulation. The latency of the presynaptic volley showed an initial decrease in a minimum of 5% below base-line value when [K’10 reached -4.5 mM. As [K’10 continued to rise, latency progressively increased, reaching a level, when [K’10 exceeded 9 mM, equivalent to that seen under conditions approaching conduction block during repetitive stimulation (for comparison, see Fig. 3). Amplitude varied in concert with latency, initially showing a slightly elevated phase (2-3%), then declining to 10% of baseline amplitude. The level of [K’10 at which changing latency and amplitude again equaled their base-line values is the same for both, -6 mM [5.6 t 0.6 (SD) mM, YI = 31. This represented an equilibrium point between the positive and negative excitability changes produced by elevated [K+], . Although artificial elevation of [K’10 yielded biphasic excitability changes in the presynaptic afferents that were similar to those seen during repetitive stimulation, several discrepancies were noted, particularly with regard to the period of increased excitability. First, the peak magnitude of the period of increased conduction velocity (4.5 t 0.5%) brought on by superfused K+ was about half the peak value seen during repetitive stimulation (8.1 t 0.9%; significant difference by paired t test at P < 0.00 1). Also, the corresponding small increase in amplitude of the fiber volley during K+ superfusion was not consistently observed during repetitive stimulation. DISCUSSION
Relative contribution of presynaptic and postsynaptic sources to activitydependent rises in [K+], Activity-evoked increases in [K’10 in the stratum radiatum of the CA 1 region of the rat hippocampus were investigated to determine their source and effects upon the excitability of CA 1 afferents. Activity-evoked K+ efflux in the stratum radiatum of CA1 is likely to be derived from afferent axons, pyramidal cell dendrites, and inhibitory interneurons. Spike initiation throughout the perikaryon of the pyramidal cell has been shown to be a large source of K+ (6), but it is unlikely that the [K’10 rises in the stratum pyramidale appre-
AND
EXCITABILITY
413
ciably extend to the stratum radiatum (2). In these experiments we sought to differentiate between presynaptic and postsynaptic sources of K+ and did not attempt to identify the conductances or cell types that contributed to the postsynaptic K+ efflux. Abolition of postsynaptic activity by any of several means yields a dramatic decrease in activity-evoked rises in [K+],. Blockade of postsynaptic activity was achieved through pharmacologic actions at both presynaptic and postsynaptic sites via receptor-mediated and non-receptor-mediated mechanisms. Since the large reduction in activity-evoked increases in [rc’], is independent of the method used to block synaptic transmission and is readily reversible in normal solution, it can be concluded that the majority of activity-evoked potassium release derives from postsynaptic elements. The rise in [K’10 seen during postsynaptic blockade can therefore be attributed to the presynaptic fibers alone, assuming that postsynaptic blockade is complete. The contribution of the postsynaptic elements to increases in [K+], can be estimated by subtracting the amount attributable to the fibers alone from the total rise seen with intact postsynaptic activity. Since postsynaptic blockade may not be complete, measures of [K’10 during postsynaptic blockade may overestimate the rise in [K+], attributable to the fibers alone; this possibility, however, renders more conservative the estimates of the predominance of postsynaptic contributions to activity-evoked increases in [K+], . The ratio of pre- to postsynaptically derived increases in [K’10 seen at low frequencies or for single stimuli is difficult to evaluate given the small quantity of K+ released from the fibers; this figure has been placed at - 1: 1 for single shocks (2). However, repetitive stimulation at increasingly higher frequencies yields a steep rise in [K’10 attributable to postsynaptic elements, thus producing a highly asymmetrical ratio of pre- to postsynaptic contribution to the total rise in [K+],. This ratio typically ranges from 1:4 at 5 Hz to 1:20 at 40 Hz, narrowing somewhat at frequencies above 40 Hz. At all frequencies above 5 Hz examined here, greater than 75% of activity-evoked K+ efflux derived from postsynaptic elements. These results indicate that the relative contributions of presynaptic and postsynaptic sources to activity-evoked rises in [K’10 are stimulation frequency dependent. When
414
N. P. POOLOS,
M. D. MAUK,
postsynaptic activity is abolished, the rise in [K], with increasing stimulation frequency is approximately linear, implying that the presynaptic fibers release a nearly constant quantity of Kt per impulse; in contrast, when postsynaptic activity is intact, the rise in [K’10 attributable to postsynaptic elements shows a sigmoidal rise with increasing stimulation frequency up to 40 Hz, followed by a gradual decline at higher frequencies. Increased postsynaptic activity was observed in the initial responses to repetitive stimulation (as can be seen in the traces of Fig. 5A) and is presumably responsible for the increased K+ efflux. Increased postsynaptic activity may represent the phenomena of paired-pulse and “frequency” facilitation, which have been observed with repetitive stimulation of CA 1 afferents (7,33).
Excitability changes resultingfrom elevation of [K+], Because potassium is the principal ion determining resting potential, [K+], elevated above its base-line level in the extracellular space (ECS) of 3 mM affects excitability through depolarization of neuronal membranes (1, 11, 3 1). As shown by superfusion of solutions containing elevated [K+], increases in [K+], up to 6 mM produce an increase in excitability, as is evident in the increased conduction velocity of the presynaptic volley. This is most likely due to a small depolarization, which reduces resting potential toward threshold. With [K’10 increased beyond 6 mM, excitability progressively diminishes, presumably from increased inactivation of voltagedependent sodium conductances that occurs with increasing depolarization ( 14). An [K’10 of -6 mM was a consistent value at which positive and negative changes in excitability were in balance, as indicated by the equivalence of both latency variation and amplitude with their base-line values. A similar value of about 6 mM for elevated [K’10 yielding baseline excitability was found in the cerebellar cortex ( 16). It should be noted that a significant discrepancy exists between the recorded levels of [K’10 derived from stimulation and from superfusion. For a given excitability change, the corresponding peak [K’10 recorded during repetitive stimulation is less than would be expected from the superfusion data. This discrepancy may result from the tendency
AND
J. D. KOCSIS
of K+-ISMs to underestimate activity-evoked changes in the actual perineuronal [K+], (20, 23). The extracellular dead space created around the ISM tip is at least an order of magnitude larger in linear dimension than the 200 A ECS surrounding neuronal elements (24); therefore, following a finite efflux of K+ which equilibrates within the contiguous ECS, the peak concentration of K+ within the dead space is likely to be significantly less than the actual [K+], initially within the perineuronal ECS. These inaccuracies are obviated during K+ superfusion, since the bath represents a virtually infinite source of K+, which enables all extracellular spaces to be in equilibrium with respect to [K+],. The exact correspondence between activity-evoked [K’JO and excitability changes is further obscured by the effects of other processes that alter excitability, such as electrogenic pump activity. In light of these considerations, K+ superfusion data can be regarded as fairly accurately describing the relationship between [K’10 and excitability, whereas results of stimulus-evoked increases in [K+], should be interpreted only as relative measures of K+ release.
Role of [K+], in activity-dependent modulation of conduction Regardless of the exact level of [K’10 produced by activation of CA1 postsynaptic elements, the excitability of CA1 afferents is highly sensitive to increases in [K’lO. The superfusion results show that elevated [K+], below 6 mM produces an increase in conduction velocity that resembles the supernormal period (SNP) described in several CNS and peripheral nerve preparations (for a review see 32). Reduction of [K’10 by postsynaptic blockade diminishes the magnitude of the activity-evoked conduction velocity increase, but superfusion of elevated [Kf] produces a smaller maximum increase in conduction velocity than that seen with active postsynaptic elements. This suggests that [K+], plays an important but not exclusive role in the production of activityevoked conduction velocity increases seen here, and perhaps in the SNP as well. Increases in [K’10 beyond 6 mM rapidly reduce the capability of the afferents to follow even relatively low-frequency stimulation. In contrast, when postsynaptic activity is abolished and activity-evoked increases in [K’10 are minimal, the responses of the presvnaptic fibers remain
EXTRACELLULAR
POTASSIUM
more tightly clustered around the base line, preserving the temporal and spatial characteristics of the stimulation pattern imposed upon the afferents. Degradation of conduction with repetitive stimulation-although significantly less than in the presence of postsynaptic activity-does occur in the absence of postsynaptic activity, and this suggests that factors other than [K’10 must contribute to this process. This is further evidenced by two observations: first, with intact postsynaptic activity, the maximum decrease in excitability is observed at high frequencies (e.g., 80 Hz) under conditions of less than maximum [K+],; second, with blocked postsynaptic activity, substantial decreases in excitability are observed at 80 Hz with an increase in [K’10 that was less than half that observed under conditions of low-frequency stimulation (e.g., at 10 Hz with intact postsynaptic activity) which produced little conduction degradation. Activity therefore can influence axonal excitability independently of [K+],. The phenomenon of conduction degradation following repetitive activity has been observed in central and peripheral mammalian axons and has been partially attributed to the action of an electrogenic Na+-K+-ATPase triggered by the accumulation of Na+ within the axonal membrane (28). Raymond and Lettvin (29) found that excitability became incrementally depressed with each impulse in a train and that the duration of recovery varied with total “depression,” usually on the order of tens and hundreds of seconds. In the experiments reported here, a similar impulsedependent depression was observed, particularly after high-frequency stimulation and necessitated a 30- to 120-s “rest” between delivery of trains so to dissipate any residual effects of stimulation. Delineation of two types of activity-dependent effects which alter excitability clarifies the differences between the families of curves shown in Figs. 3 and 4. With abolished postsynaptic activity, the major influences upon excitability are non-K+-mediated effects, that is, impulse-dependent depression. Since the cumulative level of depression increases with the number of impulses, excitability at the end of stimulation decreases monotonically with increasing stimulation frequency. Intact postsynaptic activity yields a more variable set of excitability curves, with excitability at the end
AND
415
EXCITABILITY
of stimulation not always decreasing with increasing stimulation frequency. This presumably results from the influence of elevated [K+],, the magnitude of which varies in a complex way with stimulation frequency (Fig. 6). The phenomena described above demonstrate that activity-evoked increases in [ K’10 modulate presynaptic excitability in the CA1 region of the rat hippocampus. The predominantly postsynaptic source of potassium efflux establishes a feedback mechanism that nonsynaptically couples the activity of presynaptic and postsynaptic cells. Small activity-evoked increases in [K’10 affect the fidelity of transmission of a train of impulses along CA1 afferents by biphasically altering their relative timing and spatial extent. When repetitive activity yields levels of [K’10 substantially above 6 mM, the afferents can be rendered temporarily inexcitable through depolarization block, thereby limiting the number of impulses that can be propagated along CA1 afferents at a given frequency. Activity-evoked increases in [K+], have also been suggested to underlie alterations in pyramidal cell excitability observed during paired-pulse facilitation (7, 3 3) and epileptogenie bursting (22, 30). While these phenomena were not specifically addressed in this study, it is likely that changes in synaptic activation of pyramidal cells accompany changes in presynaptic excitability mediated by increased [K’lO, and thus may contribute to altered postsynaptic excitability during repetitive activity (10). This possibility suggests that consideration of the effects of activity-evoked rises in [K+], may be necessary for a complete understanding of hippocampal synaptic function and the role activity has in its modulation.
ACKNOWLEDGMENTS
This work was supported in part by the National Institutes of Health, the Medical Research Service of the Veterns Administration, and the C. G. Swebilius Fund for Research at Yale Medical School. N. P. Poolos was supported in part by a Public Health Service training grant. Address reprint request to: Dr. Jeffery D. Kocsis, Dept. of Neurology (LCI-707), Yale Medical School, P-0. Box 3333, New Haven, CT 06510. Received 9 October March 1987.
1986;
accepted
in final
form
9
416
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
REFERENCES 1. ADELMAN, W. J. AND FITZHUGH, R. Solutions of the Hodgkin-Huxley equations modified for potassium accumulation in a periaxonal space. Federation Proc. 34: 1322-1329, 1975. 2. AITKEN, P. G. AND SOMJEN, G. G. The sources of extracellular potassium accumulation in the CA 1 region of hippocampal slices. Brain Res. 369: 163- 167, 1986. 3. ALGER, B. E. AND TEYLER, T. J. Potassium and shortterm response plasticity in the hippocampal slice. Brain Res. 159: 239-242, 1978. 4. ANDERSEN, P. BLISS, T. V. P., AND SKREDE, K. K. Unit analysis of hippocampal population spikes. Exp. Brain Res. 13: 208-221, 1971. 5. ANDERSEN, P., SILFVENIUS, H., SUNDBERG, S. H., SVEEN, O., AND WIGSTROM, H. Functional characteristics of unmyelinated fibres in the hippocampal cortex. Brain Res. 144: 1 I- 18, 1978. 6. BENNINGER, C., KADIS, J., AND PRINCE, D. A. EXtracellular calcium and potassium changes in hippocampal slices. Brain Res. 187: 165-182, 1980. 7. CREAGER, R., DUNWIDDIE, T., AND LYNCH, G. Paired-pulse and frequency facilitation in the CA1 region of the in vitro rat hippocampus. J. Physiol. Lond. 299: 409-424, 1980. 8. DICK, E. AND MILLER, R. F. Light-evoked potassium activity in mudpuppy retina: its relation to the b-wave of the electroretinogram. Brain Res. 154: 388-394, 1978. 9. ERLANGER, J. AND GASSER, H. S. Electrical Signs of Nervous Activity. Philadelphia, PA: Univ. of Pennsylvania Press, 19 3 7. 10. ERULKAR, S. D. AND WEIGHT, F. F. Extracellular potassium and transmitter release at the giant synapse 1977. of squid. J. Physiol. Land. 266: 209-218, 11. FRANKENHAEUSER, B. AND HODGKIN, A. L. The after-effects of impulses in the giant nerve fibres of Loligo. J. Physiol. Lond. 13 1: 34 l-376, 1956. 12. GANONG, A. H., LANTHORN, T. H., AND COTMAN, C. W. Kynurenic acid inhibits synaptic and acidic amino acid-induced responses in the rat hippocampus and spinal cord. Brain Res. 273: 170-l 74, 1983. 13. GASSER, H. S. AND GRUNDFEST, H. Action and excitability in mammalian A fibers. Am. J. Physiol. 117: 113-133, 1936. A. L. AND HUXLEY, A. F. The dual effect 14. HODGKIN, of membrane potential on sodium conductance in giant axon of Loligo. J. Physiol. Lond. 116: 497-506, 1952. 15. KOCSIS, J. D., ENG, D. L., AND BHISITKUL, R. B. Adenosine selectively blocks parallel-fiber-mediated synaptic potentials in rat cerebellar cortex. Proc. Natl. Acad. Sci. USA 81: 6531-6534, 1984. R. C., AND WAXMAN, 16. KOCSIS, J. D., MALENKA, S. G. Effects of extracellular potassium concentration on the excitability of the parallel fibres of the rat cerebellum. J. Physiol. Land. 334: 225-244, 1983. L. Extracel17. KRIZ, N., SYKOVA, E., AND VYKLICKY, lular potassium changes in the spinal cord of the cat and their relation to slow potentials, active transport, and impulse transmission. J. Physiol. Lond. 249: 167182, 1975.
18. LEAO, A. A. P. Spreading depression of activity in the cerebral cortex. J. Neurophysiol. 7: 359-390, 1944. 19. LORENTE DE NO, R. Studies on the structure of the cerebral cortex. II. Continuation of the study of the ammonic system. J. Psychol. Neurol. (Lpz.) 46: 113177, 1934. 20. Lux, H. D. AND NEWER, E. The equilibration time course of [K’10 in cat cortex. Exp. Brain Res. 17: 190205, 1973. 21. MALENKA, R. C., KOCSIS, J. D., RANSOM, B. R., AND WAXMAN, S. G. Modulation of parallel fiber excitability by postsynaptically mediated changes in extracellular potassium. Science Wash. DC 2 14: 339-34 1, 1981. 22. MCCARREN, M. AND ALGER, B. E. Use-dependent depression of IPSPs in rat hippocampal pyramidal cells in vitro. J. Neurophysiol. 53: 557-571, 1985. E. A. AND ODETTE, L. L. Model of elec23. NEWMAN, troretinogram b-wave generation: a test of the K+ hy1984. pothesis. J. Neurophysiol. 5 1: 164-182, 24. NICHOLSON, C. Brain-cell microenvironment as a communication channel. In: The Neurosciences Fourth Study Program, edited by F. 0. Schmitt and F. G. Worden. Cambridge, MA: MIT Press, 1979, p. 457-476. 25. NICHOLSON, C. Dynamics of the brain cell microenvironment. Neurosci. Res. Program Bull. 18: 177322, 1980. 26. NICHOLSON, C., TEN BRUGGENCATE, G., STOCKLE, H., AND STEINBERG, R. Calcium and potassium changes in extracellular microenvironment of cat cerebellar cortex. J. Neurophysiol. 4 1: 1026- 1039, 1978. 27. PHILLIS, J. W. AND WU, P. H. The role of adenosine and its cyclic nucleotides in central synaptic transmission. Prog. Neurobiol. 16: 187-239, 198 1. 28. RANG, H. P. AND RITCHIE, J. M. On the electrogenic sodium pump in mammalian nonmyelinated nerve fibres and its activation by various external cations. J. Physiol. Lond. 196: 183-22 1, 1968. 29. RAYMOND, S. A. AND LETTVIN, J. Y. Aftereffects of activity in peripheral axons as a clue to nervous coding. In: Physiology and Pathobiology ofAxons, edited by S. G. Waxman. New York: Raven, 1978, p. 203-225. 30. RUTECKI, P. A., LEBEDA, F. J., AND JOHNSTON, D. Epileptiform activity induced by changes in extracellular potassium in hippocampus. J. Neurophysiol. 54: 1363-1374, 1985. 31. SOMJEN, G. G. Extracellular potassium in the mammalian central nervous system. Annu. Rev. Physiol. 41: 159-177, 1979. 32. SWADLOW, H. A., KOCSIS, J. D., AND WAXMAN, S. G. Modulation of impulse conduction along the axonal tree. Annu. Rev. Biophys. Bioeng. 9: 143- 179, 1980. 33. TURNER, R. W., RICHARDSON, T. L., AND MILLER, J. J. Ephaptic interactions contribute to paired pulse and frequency potentiation of hippocampal field potentials. Exp. Brain Res. 54: 567-570, 1984. 34. WESTRUM, L. E. AND BLACKSTAD, T. W. An electron microscopic study of the stratum radiatum of the rat hippocampus (regio superior, CA 1) with particular emphasis on synaptology. J. Comp. Neurol. 119: 28 l309, 1962.
in U.S.A.
Activity-Evoked Increases in Extracellular Potassium Modulate Presynaptic Excitability in the CA1 Region of the Hippocampus N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
DepartmentofNeuroZogy,Stanford University Schoolof Medicine, Stanford, California 94305; Yale Medical School,New Haven, Connecticut06510;and Palo Alto VeteransAdministration Medical Center,Palo AZto,Calzfornia 94304
to 7 mM, asrecorded in the stratum radiatum with potassium ion-sensitive microelectrodes. 1. The effects of stimulus-evoked potassium When postsynaptic activity was blocked, acreleaseon the excitability of presynaptic axons tivity-evoked rises in [K+], were reduced to were studied in the rat hippocampal slice t25% of their former value. This suggeststhat preparation. Extracellular stimulation and re- activity-evoked increasesin [K’j, derive precording in the stratum radiatum of CA1 dominantly from postsynaptic elements. yielded a characteristic field potential corre5. Superfusion of solutions containing elsponding to the compound action potential of evated [K+] produced biphasic changesin the nonmyelinated afferents and subsequent excitability of CA1 afferents that were qualipostsynaptic activation of pyramidal cells. tatively similar to those produced by repetitive 2. Repetitive stimulation ( 1 s; 2- 100 Hz) stimulation. Elevated [K’10 below 6 mM proproduced biphasic changes in the excitability duced increased excitability, whereas [K’10 of the afferents. Initial responsesshowed in- above 6 mM yielded decreasedexcitability. creased conduction velocity and variably in6. These results demonstrate that in the creased amplitude; subsequent responses CA1 region of the hippocampus, significant showed progressively decreasing conduction rises in [K’10 occur with activity and derive velocity and amplitude tending toward con- predominantly from postsynaptic elements. duction block. Decreasesin excitability were The conduction properties of CA1 afferents maximal at the end of stimulation and were are sensitive to the level of [K’lO, whether almore pronounced with higher stimulation tered artificially or by activity. These effects frequencies. may constitute a mechanism of postsynaptic 3. When synaptic transmission was abol- modulation of presynaptic conduction operished with superfusate containing elevated ating within a broad range of afferent firing [Mg2’] (6 mM) and decreased [Ca2’] (0.25 frequencies in the hippocampus. mM), kynurenic acid (1 mM), or adenosine ( 100 PM), the ability of the fibers to follow INTRODUCTION repetitive stimulation was enhanced, as indicated by a reduction in amplitude decrement The transmembrane potassium gradient is of the presynaptic volley. The decreasein con- the principal determinant of resting potential duction velocity at the end of stimulation was and therefore can profoundly influence neulessthan half that obtained with intact post- ronal excitability ( 1, 11). During neuronal activity, relatively large increasesin extracellular synaptic activity. 4. Concomitant with changes in the excit- potassium concentration ([K+],) can occur, ability of CA1 afferents, the concentration of yielding depolarization of surrounding neuextracellular potassium ([K+],) increased up rons. This results both from an abundance of SUMMARY
404
AND
CONCLUSIONS
0022-3077/87
$1 SO Copyright
0 1987 The American
Physiological
Society
EXTRACELLULAR
POTASSIUM
potassium conductances that are activated during neuronal activity and from a restricted extracellular space in which small activityevoked effluxes of K+ produce significant changes in [K’10 (25). Because of the close packing of neuronal elements in the CNS, the possibility exists that activity of one set of neurons could affect excitability within the same or separate sets of neurons via depolarization resulting from changes in [K+],. Elevations in [K’10 have been observed in pathological states in the central nervous system (for a review, see 3 1) such as seizure discharge (30) and spreading depression (18,26). Significant rises in [K’10 have also been observed following physiological levels of activity in spinal dorsal horn (17) cerebellar cortex (26) retina (8) and hippocampus (3,6). However, it has been difficult to establish a functional role for the elevated [K’10 associated with either physiological or pathological states. It has been hypothesized that activity-evoked increases in [K+], underlie the primary afferent depolarization and serve as a mechanism of presynaptic inhibition ( 17). The excitability of the presynaptic parallel fibers of the cerebellar cortex has been found to be reduced during stimulus-evoked elevation of [K+], , and since much of the K+ release derived from postsynaptic elements, it was suggested that the increase in [K+], might act as a feedback mechanism to regulate parallel fiber excitability and therefore synaptic transmission ( 16,2 1). These observations give support to the notion that the extracellular ionic environment can mediate changes in neuronal excitability and thus may be regarded as a separate neuronal “communication channel” (24). In the present study, the effects of activitydependent rises in [K+], upon presynaptic axonal excitability were investigated in the in vitro hippocampal slice preparation. The results indicate that the majority of K+ efIlux during repetitive stimulation in the stratum radiatum derives from postsynaptic elements and that the resulting rise in [K’10 significantly affects the excitability of the CA1 afferents. It is suggested that modulation of presynaptic fiber excitability by accumulation of extracellular potassium from postsynaptic activity may constitute a feedback mechanism that limits afferent firing frequency and affects subsequent postsynaptic activation of hippocampal nvramidal cells.
AND
EXCITABILITY
405
METHODS
Female Wistar rats (Simonsen) were deeply anesthetized with pentobarbital sodium (60 mg/kg). After rapid exsanguination and craniotomy, a block of tissue was excised from one cerebral hemisphere using transverse cuts angled rostrally - 30° from the coronal plane. The tissue was secured to a small aluminum base using cyanoacrylate glue and buttressed against an agar block. Tissue slices 400 ,urn thick were cut using a Vibratome (Oxford Instruments), and the hippocampus was dissected free. The slices were then incubated in 34OC Krebs’ solution containing (in mM): (NaCl, 124; KC1 3.0; MgC12, 2.0; CaC12, 2.0; NaH,P04, 1.3; NaHC03, 26; dextrose, 10; pH 7.4; saturated with 95% 02 and 5% COJ for at least 1 h. Solutions with elevated [K] were osmotically balanced by subtraction of equimolar amounts of Na? During recording, slices were maintained in a submersion-type chamber with a volume of 1 ml through which Krebs’ flowed at -5 ml/min. A tungsten microelectrode (tip exposure 300 pm) was used for stimulation. Constant-current pulses controlled by a digital timing device were delivered by stimulus isolation units grounded through the bath. K+ ion-sensitive microelectrodes (IS+-ISMs), which allowed the recording of [K’lo and extracellular field potentials, were constructed from theta capillary glass (R & D Scientific Glass) using the methods described by Lux and Neher (20). Tip diameters were typically l-2 pm. The electrode barrel to serve as the K+-sensing electrode was filled at the tip with a 500~pm column of K+ ion exchange resin (Dow-Corning 4773 17) and backfilled with 2 M KCl; the reference barrel for recording field potentials contained 150 mM NaCl. The reference and K+ signals were led by Ag-AgCl wires to a highimpedance differential amplifier (Axon Instruments Axoprobe- 1). Field potentials from the reference barrel were independently amplified, low-pass filtered at 10 kHz, digitized (Nicolet 1170 and Neurodata DR-484) and stored on magnetic tape. The K+ potential was obtained by differentially amplifying the reference and K+ signals to eliminate common-mode potentials unrelated to K+ activity, low-pass filtering at 1 kHz, and monitoring on a chart recorder (Gould 220). A calibration curve for each K+-ISM was derived as a single exponential using points taken as a mean of measurements before and after experimentation from the following two test solutions (3 mM K+; 150 mM Na+) and 30 mM K+; 150 mM Na+). Calibration with a greater number of test solutions showed that the less time-consuming method allowed the estimation of [K’10 with an approximate error of t5%. Stimulating and recording electrodes were positioned in the stratum radiatum as shown in Fig. IA. Afferents in the stratum radiatum consist of the
406
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
Pl
11N2
v Nl
C
LO mV I
gc
FIG. 1. Characteristic field potentials evoked by stimulation in the stratum radiatum in the presence and absence of synaptic transmission. A: potassium ion-sensitive microelectrodes positioned in the stratum radiatum of CA1 allowed simultaneous recording of field potentials and activity-evoked extracellular K+ concentration rises following repetitive stimulation in the stratum radiatum. B: field potential recorded in the stratum radiatum resulting from a single stimulus. Nl corresponds to the presynaptic volley, whereas N2 and N3 reflect postsynaptic activity. C: field potential recorded in the stratum pyramidale resulting from a single stimulus in the stratum radiatum. The large negativity represents the pyramidal cell population spike. D-F: field potentials after blockade of synaptic transmission (arrow) by reduced [Ca2’] (0.25 mM) and elevated [Mg2+] (6 mM), kynurenic acid ( 1 mM), and adenosine ( 100 pm), respectively. Field potentials with intact postsynaptic activity are superimposed for comparison. Calibration applies to B-F.
Schaffer collaterals and commisural fibers, both of which are fine caliber, nonmyelinated axons with an average diameter of ~0.1 pm (19, 34). Activation of these axons leads to monosynaptic excitation of pyramidal cell apical dendrites. A single stimulus elicits a characteristic field potential (Fig. 1B) corresponding to the population presynaptic fiber volley and subsequent excitatory postsynaptic potential (EPSP). The initial positive-negative-positive (P lNl-P2) components of the field potential correspond to the source-sink-source currents of the propagating presynaptic volley, with N 1 representing the inward current of the action potential. Following Nl, a second negativity (N2) represents in-
ward currents associated with the population EPSP occurring within the pyramidal cell dendritic arbor (5). With increasing stimulation current, a positive component P3-N3 becomes superimposed upon N2; this component represents the source current deriving from action potentials in the pyramidal cell body, or the “population spike,” as shown by a simultaneous recording made in the stratum pyramidale (Fig. 1C). In the hippocampal slice, CA1 afferents course orthogonally to the pyramidal cell body apical dendrites (5). The stimulating and recording electrodes were positioned along the path of the afferents with a longitudinal separation of 0.1-0.5 mm to maxi-
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
2.0 msec
NS
low Ca
aden
.m v 101
kYn
FIG. 2. Effects of various synaptic blockers upon activity-dependent changes in the presynaptic volley. A: raster display of successive responses evoked by stimulation at 20 Hz for 1 s. Arrow marks first response of train. From Ze@ to right, responses were obtained in normal solution, reduced [Ca2’] (0.25 mM) with elevated [Mg2’] (6 mM), adenosine ( 100 pm), and kynurenic acid ( 1 mM). The presynaptic volley shows decreased excitability at the end of stimulation as evidenced by a decreased amplitude and increased latency. These decreases are minimal when postsynaptic activity is blocked. B: superimposed first (arrow) and last responses, which were hand traced from records of the stimulation trains in A.
mize the amplitude of the presynaptic volley (N 1) while preserving a sufficiently long conduction distance. The stimulus was adjusted to obtain a justmaximal postsynaptic (N2) response; this stimulus intensity was usually substantially submaximal with regard to the Nl response. Typically this resulted in a field potential with Nl and N2 amplitudes of 6 and 2 mV, respectively. Slices with a maximal synaptic potential of < 1 mV were discarded. Several agents were used to block synaptic activation of postsynaptic elements (Fig. 1, D-F). Increased [Mg2+] (6 mM) with reduced [Ca2’] (0.25 mM) blocks Ca2+- mediated transmitter release while minimally affecting presynaptic axon excitability. Kynurenic acid (1 mM) has been shown to antagonize the response of postsynaptic glutamate receptors at several sites in the hippocampus (12). Adenosine ( 100 PM) inhibits synaptic transmission through a putative presynaptic site of action (15, 27). Note that after disruption of synaptic transmission, only the presynaptic component of the response (P 1-N 1-P2) remained. Using synaptic blockers with diverse mechanisms of action enabled the identification of effects due to the absence of postsynaptic activity rather than to alterations in membrane properties or conductances specific to a given blocker. Excitability of an axon by definition varies inversely with its stimulation threshold (9). However, due to the impracticality of obtaining single axon recordings of fine caliber CA 1 afferents, conduction
velocity and amplitude of the population response were used as indirect measures of mean excitability. The correlation of conduction velocity and excitability has been established by single axon studies in which conduction velocity has been found to covary with threshold (29). When change in the latency of a response varies linearly with the conduction distance, latency variation can be used as an accurate measure of change in conduction velocity. Latency variation and change in conduction velocity will be used interchangeably here when that condition has been empirically met. Amplitude of a population response has also been used as a measure of excitability ( 13), and as it represents the sum of currents from synchronously activated axons, it correlates well with the number of active units (4). However, variations in resting potential will result in varying amplitudes of the constituent action currents, making the amplitude of the population response only an approximation of the number of active axons. Substantial reductions in the amplitude of the fiber volley reflect the occurrence of conduction block in a large fraction of the axon population. Evaluation of both latency variation and amplitude yields more information about excitability changes than either one alone, with latency variation giving a measure of the average excitability of a population of stimulated afferents, and amplitude approximating the size of the population.
408
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
A
A
A
050
A cl
A
A
A
0
I
0
8
"60 n 20
0
I 0
0
i A
0
0
Ofi 8A
l 1
I
I
0.4 TIME AFTER STIMULATION
0 A
A
0
1
1
0.6
I
I
0.8
I
1.0
ONSET (set)
0
8
.lO
A Cl
8
l
A
0
8
.20
0 0
A
A
A
0 0
A
20
A
050 ugj
A A
A
A
A
A80
0
1
0.0
I
I
I
0.2
I
I
0.4
I
I
0.6
TIME AFTER STIMULATION
I
OS
I
J
1.0
ONSET (set)
FIG. 3. Changes in excitability of the CA 1 afferents during repetitive stimulation with intact postsynaptic activity. Data shown in Figs. 3 and 4 are from a single representative experiment. A: change in latency of the presynaptic volley as a percentage of base-line value (first response in train) for responses recorded at various intervals during 1 s of stimulation at frequencies ranging from 10 to 80 Hz. B: amplitude of the presynaptic volley as a percentage of base-line value during 1 s of stimulation at same frequencies as in A.
RESULTS
Changes in presynapticjiber excitability produced by repetitive stimulation Activity-evoked changes in the excitability of CA1 afferents were studied by delivering trains of stimuli to the stratum radiatum. Records of successivefield potentials evoked by l-s stimulation trains at 20 Hz in normal Krebs’ solution and under conditions of synaptic blockade are shown in Fig. 24 superimpositions of the first (arrow) and last responsesof these trains are shown in Fig. 2B. In normal solution (NS), with intact postsyn-
aptic activity, significant changes in the amplitude and latency of the presynaptic volley occurred during repetitive stimulation. Two changeswere particularly evident when comparing the first and last responsesof a train: the amplitude of the last presynaptic volley had decreasedvirtually to the point of extinction, and its latency, as measured from the point of peak amplitude, had substantially increased.The large change in both of these parameters over the course of stimulation indicatesdecreasingexcitability of the presynaptic axons during repetitive stimulation. The excitability of the afferents rapidly approached
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
409
A
~80 060
I
1
1
I
I
0.6
0.4 TIME AFTER STIMULATION
ONSET (set)
A
A8O
20
0
1
0.0
I
I
0.2
I
I
I
0.4
I
0.6
TIME AFTER STIMULATION
I
I
0.8
I
1
1.0
ONSET (set)
FIG. 4. Changes in excitability of the CA 1 afferents during repetitive stimulation after abolition of postsynaptic activity with reduced [Ca”] (0.25 mM) and elevated [Mg2’] (6 mM). Data shown in Figs. 3 and 4 are from a single representative experiment. A: change in latency of the presynaptic volley as a percentage of base-line value (first response in train) for responses recorded at various intervals during 1 s of stimulation at frequencies ranging from 10 to 80 Hz. B: amplitude of the presynaptic volley as a percentage of base-line value during 1 s of stimulation at same frequencies as in A.
conduction block-that is, the population of axons responsive to a constant stimulus intensity progressively declined, and those still capable of conduction did so with decreased conduction velocity. Blockade of synaptic activation of pyramidal cell dendrites by CA1 afferents dramatically reduced the decline in excitability of the afferents during repetitive stimulation. As shown in Fig. 2, postsynaptic activity was blocked by several agents with differing mechanisms and sites of action. Regardless of the agent used, reduction of postsynaptic activity led to smaller decreases in amplitude and smaller increases in latency during repetitive
stimulation (as is evident by comparison of the first and last responses in a train) and thus greatly lessened the extent of conduction block during repetitive stimulation. Although the degree of postsynaptic blockade varied with the agent used, all were effective in restricting excitability changes of the presynaptic fibers during repetitive stimulation. When measured with single stimuli, base-line amplitude and conduction velocity of the presynaptic volley were not significantly affected by bath application of blocking agents. Quantitative comparisons of presynaptic fiber excitability during repetitive stimulation through a range of frequencies (Fig. 3) dem-
410
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
Bi 1 : :
C
k
LX
NS
2.0
40 mV msec
low ca
7
D 5
[K+1 0
(mM) 3 FIG. 5. Rises in extracellular K+ concentration ([K+],) following repetitive stimulation with intact and abolished postsynaptic activity. A: raster display of successive responses evoked by stimulation for 1 s at 20 Hz with intact postsynaptic activity. Arrow marks first response of train. B: repetitive stimulation as in A following abolition of postsynaptic activity with reduced [Ca2’] (0.25) and elevated [Mg2’] (6 mM). C: rise in [K’10 recorded during responses to stimulation shown in A. D: rise in [K’10 recorded during responses to stimulation shown in B.
onstrated the variation of excitability changes with stimulation frequency and revealed a transitory phaseof increased fiber excitability. Repetitive stimulation for 1 s at various frequencies throughout the range of lo-80 Hz produced biphasic changes in the excitability of the presynaptic fibers: several responsesfollowing the first showed decreasedlatency and constant or slightly increased amplitude of the fiber volley, whereas subsequent responses showed progressively increasing latency and decreasing amplitude. The latency variations increasedlinearly with longitudinal separation of the stimulating and recording electrodes (data not shown), thus represented changesin the conduction velocity of the fiber volley. The duration of the period of increasedconduction velocity varied approximately inversely with the stimulation frequency, but the peak magnitude of the conduction velocity increase was constant for all frequencies, - 8%. The subsequent period of decreasing conduction velocity and amplitude yielded, at the end of stimulation, a net decrease in both of these parameters from base-line value. This net decreasein excitability was observed for ah frequencies from 10 to 80 Hz, with its magnitude
varying in a complex way with frequency. The magnitudes of both increasesand decreasesin excitability showed small variability among four experiments: the mean standard deviations of measurements of latency and amplitude for all frequencies in normal solution were 5.7 and 10.9%, respectively. Both latency and amplitude variations during repetitive stimulation were greatly reduced when postsynaptic activity was blocked. Figure 4 shows the effects on excitability following blockade of postsynaptic activity by reduced [Ca2’] (0.25 mM) and elevated [Mg2’] (6 mM). Biphasic latency shifts were present, but the magnitudes of both increasesand decreases in latency were attenuated; similarly, amplitude declined over the course of stimulation, but the net change at the end of stimulation was a fraction of the value seen with intact postsynaptic activity.The differences between excitability changesobserved with and without postsynaptic activity were most marked at stimulation frequencies of 40 Hz and higher. At 40 Hz, the mean decrease in conduction velocity at the end of stimulation with blocked postsynaptic activity was 6.7 t 1.7% (SD), y2= 4, compared with a 22.9 t 4.5% decrease
EXTRACELLULAR
POTASSIUM
AND
EXCITABILITY
411
.
A7
.
5
5
10
60
I
[K+1 0
(mM)
3
:[email protected]
1K+1 0
0 0
(mM)3 t
80 ,
O ,
2
10
,
0
*
20
0
,
40 STIMULATION
0
,
1
60
0
0
0
,
,
80
,
,
loo
FREQUENCY (Hz)
FIG. 6. Rises in extracellular K+ concentration ([K],) during repetitive stimulation with intact and abolished postsynaptic activity. A: rises in [K+], recorded with intact postsynaptic activity during 1 s of repetitive stimulation at frequencies ranging from 5 to 60 Hz. B: rises in [K+], recorded after abolition of postsynaptic activity with reduced [Ca”] (0.25) and elevated [Mg2’] (6 mM) during 1 s of repetitive stimulation at same frequencies as in A. Calibration of [K+], in A also applies to B. C: rises in [K+], during 1 s of repetitive stimulation at frequencies ranging from 2 to 100 Hz with intact (JiZZedcircles) and abolished (open circles) postsynaptic activity. Data shown are from a single representative experiment.
with intact postsynaptic activity; similarly, the decline in amplitude in the absence of postsynaptic activity was 3.0 t 7.2% versus 55.8 t 16.7% in the presenceof postsynaptic activity. Evaluation of paired t test statistics found significant differences between excitability changeswith intact and abolished postsynaptic activity at 20 Hz (P < 0.05) and 40-80 Hz (P c 0.001). These results were similar with blockade of synaptic transmission by either reduced [Ca”‘] (0.25 mM) and elevated [Mg2’] (6 mM) or by application of kynurenic acid (1 mM) or adenosine (100 PM) and were completely reversible in normal solution.
Change.s in [K’jo from presynaptic and postsynaptic sources Potassium ion-sensitive microelectrodes were used to correlate excitability changes of the presynaptic fibers with simultaneous changes in [K’10 in the stratum radiatum. A
single stimulus with intact postsynaptic activity evoked a rise in [K’10 of -0.1 mM, but, as will be discussedbelow, considerations of microelectrode dead spacemake this value and those obtained during repetitive stimulation likely significant underestimates of actual [K+],. Repetitive stimulation for 1 s through a range of frequencies up to 100 Hz increased [K+], up to 7 mM, a rise of 4 mM from the baseline of 3 mM. Peak stimulus-evoked [K’10 tended to occur with stimulation at 40 Hz. Decay of the rise in [K’10 occurred exponentially with a time constant of 1.1 s with intact postsynaptic activity and 1.5 s with blocked postsynaptic activity. Increasesin [K’10 above 6 mM usually decayed with an undershoot. Repetitive stimulation for 1 s at 20 Hz produced an increase in [K’10 to over 6 mM (Fig. 5, A and C). Blockade of postsynaptic activity led to a profound reduction of activity-evoked [K’lO, yielding an increase of only 0.2 mM at
412
N. P. POOLOS,
4
t
56
7
100
M. D. MAUK,
AND
J. D. KOCSIS
’ E’l, (mW 200
12 mM K+
TIME
(set)
FIG. 7. Changes in excitability of CA 1 afferents during superfusion of Krebs’ solution containing elevated [K+]. Data shown are from a single representative experiment. Osmotically balanced Krebs’ with 12 mM K+ was superfused into the bath, and single stimuli applied every 6 s. Changes in latency (Jilled circles) and amplitude (open squares) of the presynaptic volley are shown as percentages of the base-line value. Changes in extracellular K+ concentration ([K+],) over time were recorded with a K+-ISM. The initially unchanging value of [K+], reflects the transit time of the superfusate into the bath. At a time beyond that shown here, [K’10 approached 12 mM and the amplitude of N 1 approached zero; due to the decreasing magnitude of N 1 amplitude measurements above noise level, measurements when [K’10 exceeded 9 mM are not shown.
20 Hz (Fig. 5, B and D). As shown in Fig. 6A, when a range of stimulation frequencies were used in normal solution, activity-evoked K+ release increased rapidly with increasing frequency up to 40 Hz; above 40 Hz, peak [K’10 levels gradually declined with increasing stimulation frequency. Blockade of postsynaptic activity reduced activity-evoked [K+], risesto a small fraction of their former magnitude as shown in Fig. 6, B and Cusing reduced [Ca2’] and elevated [Mg2’] as a synaptic blocker. The ratio of peak [K+], with postsynaptic activity intact and abolishedwas pooled over the range of frequencies 5-80 Hz; the mean of this value over three experiments was 8.13 t 4.92 (SD). In all experiments, with stimulation of 10 Hz or above, the ratio of peak stimulus-evoked [K’10 arising from intact and abolished postsynaptic activity was at least 3: 1. This result was independent of the agent used to block postsynaptic activity and thus suggeststhat activity-evoked increasesin [K’10 in the stratum radiatum can be principally attributed to
activity of the postsynaptic elements. Since during postsynaptic blockade the observed reduction in activity-evoked [K’10 coincided with the restriction of presynaptic fiber excitability changes, the correlation between elevated [K’10 and presynaptic excitability was investigated through artificial elevation of [K+], .
Changes in CA1 aferent excitability induced by superfusion of Krebs’ containing elevated [K’jO To assess the effects of elevated [K’10 upon the excitability of the presynaptic afferents, osmotically balanced Krebs’ solution containing 12 mM K+ was superfused into the bath, causing [K’10 in the region of the recording K+-ISM to rise gradually. During the rise in [K’lO, single stimuli were delivered every 6 s and changesin latency and amplitude of the fiber volley were measured. The results of this experiment are depicted in Fig. 7. Elevating [K+], by bath superfusion induced
EXTRACELLULAR
POTASSIUM
biphasic changes in the conduction properties of CA 1 afferents which were qualitatively similar to those induced by repetitive stimulation. The latency of the presynaptic volley showed an initial decrease in a minimum of 5% below base-line value when [K’10 reached -4.5 mM. As [K’10 continued to rise, latency progressively increased, reaching a level, when [K’10 exceeded 9 mM, equivalent to that seen under conditions approaching conduction block during repetitive stimulation (for comparison, see Fig. 3). Amplitude varied in concert with latency, initially showing a slightly elevated phase (2-3%), then declining to 10% of baseline amplitude. The level of [K’10 at which changing latency and amplitude again equaled their base-line values is the same for both, -6 mM [5.6 t 0.6 (SD) mM, YI = 31. This represented an equilibrium point between the positive and negative excitability changes produced by elevated [K+], . Although artificial elevation of [K’10 yielded biphasic excitability changes in the presynaptic afferents that were similar to those seen during repetitive stimulation, several discrepancies were noted, particularly with regard to the period of increased excitability. First, the peak magnitude of the period of increased conduction velocity (4.5 t 0.5%) brought on by superfused K+ was about half the peak value seen during repetitive stimulation (8.1 t 0.9%; significant difference by paired t test at P < 0.00 1). Also, the corresponding small increase in amplitude of the fiber volley during K+ superfusion was not consistently observed during repetitive stimulation. DISCUSSION
Relative contribution of presynaptic and postsynaptic sources to activitydependent rises in [K+], Activity-evoked increases in [K’10 in the stratum radiatum of the CA 1 region of the rat hippocampus were investigated to determine their source and effects upon the excitability of CA 1 afferents. Activity-evoked K+ efflux in the stratum radiatum of CA1 is likely to be derived from afferent axons, pyramidal cell dendrites, and inhibitory interneurons. Spike initiation throughout the perikaryon of the pyramidal cell has been shown to be a large source of K+ (6), but it is unlikely that the [K’10 rises in the stratum pyramidale appre-
AND
EXCITABILITY
413
ciably extend to the stratum radiatum (2). In these experiments we sought to differentiate between presynaptic and postsynaptic sources of K+ and did not attempt to identify the conductances or cell types that contributed to the postsynaptic K+ efflux. Abolition of postsynaptic activity by any of several means yields a dramatic decrease in activity-evoked rises in [K+],. Blockade of postsynaptic activity was achieved through pharmacologic actions at both presynaptic and postsynaptic sites via receptor-mediated and non-receptor-mediated mechanisms. Since the large reduction in activity-evoked increases in [rc’], is independent of the method used to block synaptic transmission and is readily reversible in normal solution, it can be concluded that the majority of activity-evoked potassium release derives from postsynaptic elements. The rise in [K’10 seen during postsynaptic blockade can therefore be attributed to the presynaptic fibers alone, assuming that postsynaptic blockade is complete. The contribution of the postsynaptic elements to increases in [K+], can be estimated by subtracting the amount attributable to the fibers alone from the total rise seen with intact postsynaptic activity. Since postsynaptic blockade may not be complete, measures of [K’10 during postsynaptic blockade may overestimate the rise in [K+], attributable to the fibers alone; this possibility, however, renders more conservative the estimates of the predominance of postsynaptic contributions to activity-evoked increases in [K+], . The ratio of pre- to postsynaptically derived increases in [K’10 seen at low frequencies or for single stimuli is difficult to evaluate given the small quantity of K+ released from the fibers; this figure has been placed at - 1: 1 for single shocks (2). However, repetitive stimulation at increasingly higher frequencies yields a steep rise in [K’10 attributable to postsynaptic elements, thus producing a highly asymmetrical ratio of pre- to postsynaptic contribution to the total rise in [K+],. This ratio typically ranges from 1:4 at 5 Hz to 1:20 at 40 Hz, narrowing somewhat at frequencies above 40 Hz. At all frequencies above 5 Hz examined here, greater than 75% of activity-evoked K+ efflux derived from postsynaptic elements. These results indicate that the relative contributions of presynaptic and postsynaptic sources to activity-evoked rises in [K’10 are stimulation frequency dependent. When
414
N. P. POOLOS,
M. D. MAUK,
postsynaptic activity is abolished, the rise in [K], with increasing stimulation frequency is approximately linear, implying that the presynaptic fibers release a nearly constant quantity of Kt per impulse; in contrast, when postsynaptic activity is intact, the rise in [K’10 attributable to postsynaptic elements shows a sigmoidal rise with increasing stimulation frequency up to 40 Hz, followed by a gradual decline at higher frequencies. Increased postsynaptic activity was observed in the initial responses to repetitive stimulation (as can be seen in the traces of Fig. 5A) and is presumably responsible for the increased K+ efflux. Increased postsynaptic activity may represent the phenomena of paired-pulse and “frequency” facilitation, which have been observed with repetitive stimulation of CA 1 afferents (7,33).
Excitability changes resultingfrom elevation of [K+], Because potassium is the principal ion determining resting potential, [K+], elevated above its base-line level in the extracellular space (ECS) of 3 mM affects excitability through depolarization of neuronal membranes (1, 11, 3 1). As shown by superfusion of solutions containing elevated [K+], increases in [K+], up to 6 mM produce an increase in excitability, as is evident in the increased conduction velocity of the presynaptic volley. This is most likely due to a small depolarization, which reduces resting potential toward threshold. With [K’10 increased beyond 6 mM, excitability progressively diminishes, presumably from increased inactivation of voltagedependent sodium conductances that occurs with increasing depolarization ( 14). An [K’10 of -6 mM was a consistent value at which positive and negative changes in excitability were in balance, as indicated by the equivalence of both latency variation and amplitude with their base-line values. A similar value of about 6 mM for elevated [K’10 yielding baseline excitability was found in the cerebellar cortex ( 16). It should be noted that a significant discrepancy exists between the recorded levels of [K’10 derived from stimulation and from superfusion. For a given excitability change, the corresponding peak [K’10 recorded during repetitive stimulation is less than would be expected from the superfusion data. This discrepancy may result from the tendency
AND
J. D. KOCSIS
of K+-ISMs to underestimate activity-evoked changes in the actual perineuronal [K+], (20, 23). The extracellular dead space created around the ISM tip is at least an order of magnitude larger in linear dimension than the 200 A ECS surrounding neuronal elements (24); therefore, following a finite efflux of K+ which equilibrates within the contiguous ECS, the peak concentration of K+ within the dead space is likely to be significantly less than the actual [K+], initially within the perineuronal ECS. These inaccuracies are obviated during K+ superfusion, since the bath represents a virtually infinite source of K+, which enables all extracellular spaces to be in equilibrium with respect to [K+],. The exact correspondence between activity-evoked [K’JO and excitability changes is further obscured by the effects of other processes that alter excitability, such as electrogenic pump activity. In light of these considerations, K+ superfusion data can be regarded as fairly accurately describing the relationship between [K’10 and excitability, whereas results of stimulus-evoked increases in [K+], should be interpreted only as relative measures of K+ release.
Role of [K+], in activity-dependent modulation of conduction Regardless of the exact level of [K’10 produced by activation of CA1 postsynaptic elements, the excitability of CA1 afferents is highly sensitive to increases in [K’lO. The superfusion results show that elevated [K+], below 6 mM produces an increase in conduction velocity that resembles the supernormal period (SNP) described in several CNS and peripheral nerve preparations (for a review see 32). Reduction of [K’10 by postsynaptic blockade diminishes the magnitude of the activity-evoked conduction velocity increase, but superfusion of elevated [Kf] produces a smaller maximum increase in conduction velocity than that seen with active postsynaptic elements. This suggests that [K+], plays an important but not exclusive role in the production of activityevoked conduction velocity increases seen here, and perhaps in the SNP as well. Increases in [K’10 beyond 6 mM rapidly reduce the capability of the afferents to follow even relatively low-frequency stimulation. In contrast, when postsynaptic activity is abolished and activity-evoked increases in [K’10 are minimal, the responses of the presvnaptic fibers remain
EXTRACELLULAR
POTASSIUM
more tightly clustered around the base line, preserving the temporal and spatial characteristics of the stimulation pattern imposed upon the afferents. Degradation of conduction with repetitive stimulation-although significantly less than in the presence of postsynaptic activity-does occur in the absence of postsynaptic activity, and this suggests that factors other than [K’10 must contribute to this process. This is further evidenced by two observations: first, with intact postsynaptic activity, the maximum decrease in excitability is observed at high frequencies (e.g., 80 Hz) under conditions of less than maximum [K+],; second, with blocked postsynaptic activity, substantial decreases in excitability are observed at 80 Hz with an increase in [K’10 that was less than half that observed under conditions of low-frequency stimulation (e.g., at 10 Hz with intact postsynaptic activity) which produced little conduction degradation. Activity therefore can influence axonal excitability independently of [K+],. The phenomenon of conduction degradation following repetitive activity has been observed in central and peripheral mammalian axons and has been partially attributed to the action of an electrogenic Na+-K+-ATPase triggered by the accumulation of Na+ within the axonal membrane (28). Raymond and Lettvin (29) found that excitability became incrementally depressed with each impulse in a train and that the duration of recovery varied with total “depression,” usually on the order of tens and hundreds of seconds. In the experiments reported here, a similar impulsedependent depression was observed, particularly after high-frequency stimulation and necessitated a 30- to 120-s “rest” between delivery of trains so to dissipate any residual effects of stimulation. Delineation of two types of activity-dependent effects which alter excitability clarifies the differences between the families of curves shown in Figs. 3 and 4. With abolished postsynaptic activity, the major influences upon excitability are non-K+-mediated effects, that is, impulse-dependent depression. Since the cumulative level of depression increases with the number of impulses, excitability at the end of stimulation decreases monotonically with increasing stimulation frequency. Intact postsynaptic activity yields a more variable set of excitability curves, with excitability at the end
AND
415
EXCITABILITY
of stimulation not always decreasing with increasing stimulation frequency. This presumably results from the influence of elevated [K+],, the magnitude of which varies in a complex way with stimulation frequency (Fig. 6). The phenomena described above demonstrate that activity-evoked increases in [ K’10 modulate presynaptic excitability in the CA1 region of the rat hippocampus. The predominantly postsynaptic source of potassium efflux establishes a feedback mechanism that nonsynaptically couples the activity of presynaptic and postsynaptic cells. Small activity-evoked increases in [K’10 affect the fidelity of transmission of a train of impulses along CA1 afferents by biphasically altering their relative timing and spatial extent. When repetitive activity yields levels of [K’10 substantially above 6 mM, the afferents can be rendered temporarily inexcitable through depolarization block, thereby limiting the number of impulses that can be propagated along CA1 afferents at a given frequency. Activity-evoked increases in [K+], have also been suggested to underlie alterations in pyramidal cell excitability observed during paired-pulse facilitation (7, 3 3) and epileptogenie bursting (22, 30). While these phenomena were not specifically addressed in this study, it is likely that changes in synaptic activation of pyramidal cells accompany changes in presynaptic excitability mediated by increased [K’lO, and thus may contribute to altered postsynaptic excitability during repetitive activity (10). This possibility suggests that consideration of the effects of activity-evoked rises in [K+], may be necessary for a complete understanding of hippocampal synaptic function and the role activity has in its modulation.
ACKNOWLEDGMENTS
This work was supported in part by the National Institutes of Health, the Medical Research Service of the Veterns Administration, and the C. G. Swebilius Fund for Research at Yale Medical School. N. P. Poolos was supported in part by a Public Health Service training grant. Address reprint request to: Dr. Jeffery D. Kocsis, Dept. of Neurology (LCI-707), Yale Medical School, P-0. Box 3333, New Haven, CT 06510. Received 9 October March 1987.
1986;
accepted
in final
form
9
416
N. P. POOLOS,
M. D. MAUK,
AND
J. D. KOCSIS
REFERENCES 1. ADELMAN, W. J. AND FITZHUGH, R. Solutions of the Hodgkin-Huxley equations modified for potassium accumulation in a periaxonal space. Federation Proc. 34: 1322-1329, 1975. 2. AITKEN, P. G. AND SOMJEN, G. G. The sources of extracellular potassium accumulation in the CA 1 region of hippocampal slices. Brain Res. 369: 163- 167, 1986. 3. ALGER, B. E. AND TEYLER, T. J. Potassium and shortterm response plasticity in the hippocampal slice. Brain Res. 159: 239-242, 1978. 4. ANDERSEN, P. BLISS, T. V. P., AND SKREDE, K. K. Unit analysis of hippocampal population spikes. Exp. Brain Res. 13: 208-221, 1971. 5. ANDERSEN, P., SILFVENIUS, H., SUNDBERG, S. H., SVEEN, O., AND WIGSTROM, H. Functional characteristics of unmyelinated fibres in the hippocampal cortex. Brain Res. 144: 1 I- 18, 1978. 6. BENNINGER, C., KADIS, J., AND PRINCE, D. A. EXtracellular calcium and potassium changes in hippocampal slices. Brain Res. 187: 165-182, 1980. 7. CREAGER, R., DUNWIDDIE, T., AND LYNCH, G. Paired-pulse and frequency facilitation in the CA1 region of the in vitro rat hippocampus. J. Physiol. Lond. 299: 409-424, 1980. 8. DICK, E. AND MILLER, R. F. Light-evoked potassium activity in mudpuppy retina: its relation to the b-wave of the electroretinogram. Brain Res. 154: 388-394, 1978. 9. ERLANGER, J. AND GASSER, H. S. Electrical Signs of Nervous Activity. Philadelphia, PA: Univ. of Pennsylvania Press, 19 3 7. 10. ERULKAR, S. D. AND WEIGHT, F. F. Extracellular potassium and transmitter release at the giant synapse 1977. of squid. J. Physiol. Land. 266: 209-218, 11. FRANKENHAEUSER, B. AND HODGKIN, A. L. The after-effects of impulses in the giant nerve fibres of Loligo. J. Physiol. Lond. 13 1: 34 l-376, 1956. 12. GANONG, A. H., LANTHORN, T. H., AND COTMAN, C. W. Kynurenic acid inhibits synaptic and acidic amino acid-induced responses in the rat hippocampus and spinal cord. Brain Res. 273: 170-l 74, 1983. 13. GASSER, H. S. AND GRUNDFEST, H. Action and excitability in mammalian A fibers. Am. J. Physiol. 117: 113-133, 1936. A. L. AND HUXLEY, A. F. The dual effect 14. HODGKIN, of membrane potential on sodium conductance in giant axon of Loligo. J. Physiol. Lond. 116: 497-506, 1952. 15. KOCSIS, J. D., ENG, D. L., AND BHISITKUL, R. B. Adenosine selectively blocks parallel-fiber-mediated synaptic potentials in rat cerebellar cortex. Proc. Natl. Acad. Sci. USA 81: 6531-6534, 1984. R. C., AND WAXMAN, 16. KOCSIS, J. D., MALENKA, S. G. Effects of extracellular potassium concentration on the excitability of the parallel fibres of the rat cerebellum. J. Physiol. Land. 334: 225-244, 1983. L. Extracel17. KRIZ, N., SYKOVA, E., AND VYKLICKY, lular potassium changes in the spinal cord of the cat and their relation to slow potentials, active transport, and impulse transmission. J. Physiol. Lond. 249: 167182, 1975.
18. LEAO, A. A. P. Spreading depression of activity in the cerebral cortex. J. Neurophysiol. 7: 359-390, 1944. 19. LORENTE DE NO, R. Studies on the structure of the cerebral cortex. II. Continuation of the study of the ammonic system. J. Psychol. Neurol. (Lpz.) 46: 113177, 1934. 20. Lux, H. D. AND NEWER, E. The equilibration time course of [K’10 in cat cortex. Exp. Brain Res. 17: 190205, 1973. 21. MALENKA, R. C., KOCSIS, J. D., RANSOM, B. R., AND WAXMAN, S. G. Modulation of parallel fiber excitability by postsynaptically mediated changes in extracellular potassium. Science Wash. DC 2 14: 339-34 1, 1981. 22. MCCARREN, M. AND ALGER, B. E. Use-dependent depression of IPSPs in rat hippocampal pyramidal cells in vitro. J. Neurophysiol. 53: 557-571, 1985. E. A. AND ODETTE, L. L. Model of elec23. NEWMAN, troretinogram b-wave generation: a test of the K+ hy1984. pothesis. J. Neurophysiol. 5 1: 164-182, 24. NICHOLSON, C. Brain-cell microenvironment as a communication channel. In: The Neurosciences Fourth Study Program, edited by F. 0. Schmitt and F. G. Worden. Cambridge, MA: MIT Press, 1979, p. 457-476. 25. NICHOLSON, C. Dynamics of the brain cell microenvironment. Neurosci. Res. Program Bull. 18: 177322, 1980. 26. NICHOLSON, C., TEN BRUGGENCATE, G., STOCKLE, H., AND STEINBERG, R. Calcium and potassium changes in extracellular microenvironment of cat cerebellar cortex. J. Neurophysiol. 4 1: 1026- 1039, 1978. 27. PHILLIS, J. W. AND WU, P. H. The role of adenosine and its cyclic nucleotides in central synaptic transmission. Prog. Neurobiol. 16: 187-239, 198 1. 28. RANG, H. P. AND RITCHIE, J. M. On the electrogenic sodium pump in mammalian nonmyelinated nerve fibres and its activation by various external cations. J. Physiol. Lond. 196: 183-22 1, 1968. 29. RAYMOND, S. A. AND LETTVIN, J. Y. Aftereffects of activity in peripheral axons as a clue to nervous coding. In: Physiology and Pathobiology ofAxons, edited by S. G. Waxman. New York: Raven, 1978, p. 203-225. 30. RUTECKI, P. A., LEBEDA, F. J., AND JOHNSTON, D. Epileptiform activity induced by changes in extracellular potassium in hippocampus. J. Neurophysiol. 54: 1363-1374, 1985. 31. SOMJEN, G. G. Extracellular potassium in the mammalian central nervous system. Annu. Rev. Physiol. 41: 159-177, 1979. 32. SWADLOW, H. A., KOCSIS, J. D., AND WAXMAN, S. G. Modulation of impulse conduction along the axonal tree. Annu. Rev. Biophys. Bioeng. 9: 143- 179, 1980. 33. TURNER, R. W., RICHARDSON, T. L., AND MILLER, J. J. Ephaptic interactions contribute to paired pulse and frequency potentiation of hippocampal field potentials. Exp. Brain Res. 54: 567-570, 1984. 34. WESTRUM, L. E. AND BLACKSTAD, T. W. An electron microscopic study of the stratum radiatum of the rat hippocampus (regio superior, CA 1) with particular emphasis on synaptology. J. Comp. Neurol. 119: 28 l309, 1962.
